-----------------------------
Guide to replication materials for
"Intrinsic Social Incentives in State and Non-State Armed Groups"
Materials prepared by Cyrus Samii, NYU (cds2083@nyu.edu)
December 2021
-----------------------------

# Statistical software: 

Stata version 13.1

Requires the following packages to be installed from SSC:

boottest
sutex
estout

# Data files:

1. pla-civ-mixed-repdata.dta
Data from Nepal on games outcomes from sessions with PLA members only, PLA mixed with civilians, and civilians only. These data are used to replicate the main results and robustness checks for Nepal.

2. nepal-desertion-repdata.dta
Data from Nepal on desertions and killed in action for members of respondents' first units.  These data are used for a robustness check for Nepal. 

3. cdi-militia-repdata.dta
Data from Cote d'Ivoire on games outcomes from sessions with militia and military members.  These data are used to replicate the main results and robustness checks for Cote d'Ivoire.

4. kurdistan-repdata.dta
Data from Kurdistan on games outcomes from sessions with militia and military members.  These data are used to replicate the main results and robustness checks for Kurdistan.

# Data files notes:

For the following covariates in the Nepal data

- c_maoistevbw "Local Maoist attacks when joined" 
- c_stateevbw "Local state attacks when joined"
- c_maoistshbw "Share attacks Maoist when joined"

These were created using the 2016 version of the ACLED georeferenced conflict events data for Nepal. The variables measure the cumulative number of Maoist and state attacks as well as the Maoist share of total cumulative attacks within a 50km radius of respondents' home districts at the time the respondent joined. Code to reproduce these measures is available from the authors upon request.

All other variables are based on original field data collection by the authors.  The protocols for the sample selection and the games are described in the main text.  The samples of 

# Analysis scripts

master.do
This Stata do file runs all of the analyses and produces all tables and graphs, which are saved to the "tabs-figs" subdirectory.

# Replicating the analysis

To replicate the analysis, be sure to have all of the datafiles stored in your working directory, and set Stata to work from that working directory.  Also be sure that there is a directory (i.e. folder) called "tabs-figs" inside your working directory.  Then, run master.do to reproduce all of the tables and figures that appear in the paper and appendix.

# Replication notes

The wild bootstrap p-values do not always print properly in the .tex files produced from esttab.  Therefore, we save log files and manually enter the p-values into the respective .tex files. To find all of the wild bootstrap p-values, please consult the respective log files.

In Table 2, there is some very minor (second decimal place) differences in column (2) results across versions of the paper due to our using one round of predictive mean matching imputation to fill in 3 missing values for the “rankorder” variable, 8 missing values for the “combat_num” variable, and 4 missing values for the “wounded” variable.

For Figure 7, there are very slight differences in the way the points appear in the scatterplot due to randomness from the “jitter()” function in Stata.

   